Project summary

AI can make a research claim sound finished before anyone checks it. I made a one-page Falsifiability Sheet to slow that down. It puts one claim next to the exact sources used for it. First, a person checks the claim. Then AI checks only the same sources. The sheet gives one result: Green, Yellow, Red, or STOP. STOP means the needed evidence is not there. The sheet does not decide what is true, and it does not replace a trained scholar.

Alexandria University approved a guide in Arabic and English for using AI in research, including dissertations. The university has also talked about new AI tools for research. But those plans do not give researchers a simple record for each claim. The page should show whether a claim matches its sources. I now have a draft of the sheet in Arabic. Mahmoud El Tahan is an Egyptology master's researcher at Alexandria University. He has agreed to try the sheet on one small part of his dissertation work. This is work between two researchers. It is not an official university project.

With this grant, I will run a public test for three to four months. I will use eight test cases with answers set before the work begins. I will also use one real case from Alexandria. Each case gets a human check. I will run the AI step 54 times. I will use three model families and check if the answers change. I will post every result, disagreement, error, refusal, and case I remove. If the sheet fails, I will post that too.

The Falsifiability Sheet: one claim, one set of sources, two separate checks, and one saved result.

The missing piece: Alexandria has a guide for using AI in research. This test asks one basic question. Does one AI-assisted claim match the evidence used to make it?

Why fund this now - and why it fits Manifund

Alexandria now has a guide for AI use in research. The Arabic sheet is ready as a draft, and Mahmoud has agreed to try it. I do not need money to come up with the idea. I need money to test what I already built. People outside my project will help.

Public: The sheet, data, and results will be free to read and use.

Small test: A $12,000 grant is enough to test the working sheet in a real research setting.

Fast: I can finish and post the main study within four months.

Failure still matters: I will post bad results, disagreements, removed cases, and the full AI answers too.

What I have already built and what the grant pays for

The sheet already works as a set of steps. I am asking for money to test it with other people. I need to know if the steps work well enough.

Already built: Version 6A is online as a full example. It includes the sources and the first human check. It also has the AI check, final result, and test limits. I have also tried the steps with ChatGPT, Claude, and Gemini. An Arabic draft is ready. This shows that the steps can be followed. It does not prove that the sheet is reliable.

Working V6A record — Zenodo  |  Open project repository

The grant pays for what is still missing. That means test cases with known answers, outside reviewers, and new AI sessions. It also pays for one real case in Alexandria, an Arabic check, and a public record of the whole test.

Why start with ancient writing?

Egyptian hieroglyphs are already understood, so I can test the sheet against known answers. Ancient writing is a good challenge. Small details matter, and a sure-sounding claim can go too far. I am not saying one study will prove the sheet works in every field. I am asking whether it works here and whether a larger test is worth doing.

What are this project's goals? How will you achieve them?

What the $12,000 study includes

Eight planned test cases: Before the study starts, I will post each case and its expected answer. There will be two Green, two Yellow, two Red, and two STOP cases.

One real Alexandria case: Mahmoud will use the sheet on one small part of his dissertation work. Private details will be removed when needed. If that case cannot be used, I will use a public backup. I will say clearly that the real-use test did not happen.

One outside human check for each case: The reviewer will not be told the answer I expect. Mahmoud will not be the only reviewer for his own case.

Fifty-four AI runs: 9 cases x 3 model families x 2 new sessions. This lets me see if the result changes. It is still nine cases, not 54 separate cases.

Everything will be shared: the English and Arabic sheets, prompts, and data. I will share the source packets when I have permission. I will also post notes for each field and a short guide. The final release will list every run, show the review, and explain what failed.

What the $18,000 goal adds

Twelve cases in all and 72 AI runs.

Two more real or public cases.

At least one person outside my project will run a full case from start to finish.

Some cases will get a second human check. Part of the study will also be run again to see if the results hold.

You can support one part of the work

A donor can message me through Manifund to pay for one part. That could be the Arabic test, outside review, or a full repeat by someone else. It could also pay for one more public case. Researchers can suggest a small case or volunteer to review one.

Every case must follow the same posted rules. Money does not buy a Green result. A sponsor cannot pick the answer or change the sources after the test starts. A sponsor also cannot stop me from posting a bad result.

How the test works

1. Set the rules first. Before I run anything, I will post the claims, sources, and prompts. I will also post the expected results, removal rules, and final decision rules. I will use digital hashes. They show if I change a file later.

2. Keep the two checks apart. R1 is the human check. R2 is the AI check. R2 only sees the claim, sources, and blank form. It cannot search the web or see R1's answer. If the needed evidence is missing, R2 must say STOP.

3. Start with a clean session each time. I will run each case twice in ChatGPT, Claude, and Gemini. I will save the model name, date, settings, files, prompt, refusal, and full answer.

4. Compare the results. I will count every time a Red or STOP case wrongly becomes Green. I will see how often the AI runs agree. I will also show where the person and AI disagree. Last, I will see if the Arabic sheet is easy to use from start to finish.

5. Post the result even if it is bad. I will share the test rules, data, removed cases, errors, and limits. I will hold back only two things: material I cannot share and details that could name someone.

What counts as failure, and what still helps

The worst error: A known Red or STOP case ends as Green. Big changes between new sessions or between AI models also count against the sheet.

The study does not have to end with a good score to matter. The work is useful if it gives us a full record. That record must show where the sheet works and where it changes too much. It must also show when the sheet takes too much time or should not be used.

Timeline

Month-by-month plan

Month 1: Set the rules. Have a professional Arabic proofreader check the sheet. Make sure the QR code works. Hire outside reviewers. Post the test plan before the first run.

Month 2: Run the eight planned cases. Save every file and answer. Fix only small process problems allowed by the posted plan.

Month 3: Run the Alexandria case and finish the new AI sessions. Write down what was easy, what was hard, where the evidence stopped, and where the reviewers disagreed.

Month 4: Review the results. Get an outside check on how I ran the study. If the $18,000 goal is reached, repeat part of the work. Post the data, the English-Arabic kit, the guide, and the final report.

How will this funding be used?

Budget

The $12,000 minimum pays for the whole basic study. The $18,000 goal adds more cases and more outside checking.

Where the money goes

My work: set the rules, prepare cases, run tests, study results, and post the files. Minimum $5,000 | Goal $8,000

Outside R1 reviewers and second checks: Minimum $3,500 | Goal $5,000

Alexandria Arabic test: help with the case, prepare the evidence, and remove private details: Minimum $1,000 | Goal $1,000

Outside check of the study and a full run by someone else: Minimum $750 | Goal $1,500

AI model access, saved records, file storage, and online archive: Minimum $750 | Goal $1,000

Public data, notes, English-Arabic kit, final check, and sharing: Minimum $1,000 | Goal $1,500

TOTAL: Minimum $12,000 | Goal $18,000

I plan to take this as a direct Manifund grant. Echoes of the Script also has a fiscal sponsor, Fractured Atlas, if Manifund needs that route. I will list any fee or budget change before I take the money.

Who is on your team? What's your track record on similar projects?

Team

Michael Grasa - project lead and creator of the sheet

I made the Falsifiability Sheet and will lead the study. I will set the rules and prepare the cases. I will run the AI sessions, compare the results, and post the final record. Emergent Ventures gave me about $12,000 for earlier work. I shared an early form of this work at the Archaeological Survey of India. The talk was in New Delhi in September 2025. Version 6A is a full public example. I am not claiming to be the expert who decides the Egyptian answers. Trained reviewers will do that part, and I will show where they disagree.

Mahmoud El Tahan - Alexandria research partner

Mahmoud is an Egyptian master's researcher in Egyptology at Alexandria University. He studies Late Egyptian hieratic and hieroglyphic writing. He has a bachelor's degree in archaeology. He also worked for about five years as a senior archaeologist in Alexandria. He speaks Arabic and works in this field. He will try the Arabic sheet on one small part of his dissertation work. He will note what is clear and what is not. He will also show where the evidence runs out. He is helping as one researcher. He is not speaking for Alexandria University. He will not be the only R1 reviewer for his own case.

Outside reviewers

The grant pays for outside R1 reviewers who know the subject. It also pays for one person to check how I ran the study. Reviewers will not be told the result I expect. I will name each reviewer and list their background and any conflicts. I will say exactly what each person did before I post a result.

What are the most likely causes and outcomes if this project fails?

What could go wrong, and what I will do

A Red or STOP case turns Green, or answers keep changing: I will post the failure. I will fix or stop using the sheet. I will not hide the problem by averaging the answers.

My own choices could push the result: I will set and post the files before the study. I will also post the expected answers, prompts, and rules. Reviewers will not see the answers I expect.

The Alexandria case is delayed: I will use the public backup case. I will clearly say that the real-use case did not happen.

AI models change: I will save the model name, date, settings, files, prompts, and full answers. I will not claim that a private model can be copied exactly years from now.

I made the sheet, so I could be biased: Outside reviewers will check the cases. Another person will check how I ran the study. If the goal is reached, someone outside the project will run a full case.

Few people use it: I will not pretend that the sheet has been adopted. I will post the English-Arabic kit and the notes from the study. Others can then decide if another test is worth doing.

How much money have you raised in the last 12 months, and from where?

Prior funding

Emergent Ventures gave me about $12,000 during the last 12 months. That money paid for earlier work on the Indus script and the Falsifiability Sheet. The award was announced on April 18, 2026. It did not pay for this new study. I have raised no money yet for the Alexandria test.

Public links and proof

Working V6A record — Zenodo

Alexandria University AI research guide announcement

Alexandria University research-infrastructure announcement

Mahmoud El Tahan — public professional profile

British Museum EA1006 — example source object

Emergent Ventures award announcement

Open project repository

Fiscal sponsor page, if routing is needed

Important note

I am not claiming that Alexandria University backs this project. The university links only show what the university has said in public. Mahmoud is helping as an individual researcher. Before the study starts, a professional Arabic proofreader will check the sheet. I will also test its link and QR code.

STOP COPYING ABOVE THIS LINE.

After publishing

Sign the grant agreement shown on the project page and confirm that the $12,000 minimum, $18,000 goal, cause areas, and deadline display correctly.

Optional first comment - post this only after the project is live:

The full Version 6A record is already online and linked above. The $12,000 minimum pays for the whole basic study. The $18,000 goal adds more cases and more outside review. If you want to fund or review one part, please leave a comment or message me through Manifund. I will post bad results under the same rules as good ones.